It is easy to forget how rapidly stem cell science has developed over the past few years. A glance at the stem cell papers published in leading journals hints at both the pace of progress and the competition within the field. It has only been three years since Shinya Yamanaka, at Kyoto University in Japan, published his famous paper showing that differentiated mouse cells could be engineered to behave like embryonic stem (ES) cells — able to make almost every cell type in the body1. Only a handful of genes had to be introduced and overexpressed in these cells to make this possible. By June 2007, three papers had confirmed that such 'reprogrammed' mouse cells, now known as induced pluripotent stem (iPS) cells, could indeed become all body tissues2,3,4. By the end of 2007, a trio of papers was published that demonstrated the reprogramming of human cells5,6,7. The summer of 2009 has brought three papers showing that live, fertile mice can be generated from iPS cells — a further demonstration of the cells' similarity to ES cells (For references, see iPS cells make mice that make mice). And another five papers published in August 2009 identified molecular pathways that both link reprogramming to cancer and prevent reprogramming from being more efficient8,9,10,11,12.

In the wider field of stem cell research, there is optimism over the potential impact of stem cells on the study of and treatment of disease. Since 2005, at least nine new peer-reviewed journals have been launched in the fields of stem cells and regenerative medicine. But perhaps just as striking as the work accomplished so far is the effort that is still required to turn stem cells into robust tools for modelling disease, screening drugs and treating degenerative diseases. And the process of understanding how stem cells function naturally has barely begun.

Defining the cells

All stem cells are characterized by their ability to produce both more stem cells (the process of self-renewal) and cells that take on more specialized tasks (the process of differentiation). They proliferate to produce new tissue as an embryo develops and repair damaged tissue throughout an individual's life. Haematopoietic stem cells can produce patrolling macrophages, attacking T cells, antibody-producing B cells and every other sort of blood cell, including more haematopoietic stem cells. Neural stem cells produce neurons as well as the supporting glial cells. Tissue-specific stem cells, such as neural stem cells and haematopoietic stem cells, when extracted from mammalian fetal and adult tissues, can reproduce themselves and the specialized cells of a particular tissue but not, it is generally believed, cells outside that tissue type. Neural stem cells cannot make blood, for example.

By contrast, ES cells and their engineered counterparts, iPS cells, are capable of making every cell type in the body, a property known as pluripotency. ES cells are taken from the inner cell mass of a mammalian blastocyst and grown in culture in the laboratory. Unlike tissue-specific stem cells, pluripotent stem cells can expand indefinitely in culture, generating hundreds of times as many cells as the researchers started with. Using any kind of stem cell for therapy or study will probably require large numbers of them, so this property is important. The downside is that although ES cells and iPS cells can be easily expanded, they do not form specialized cells in culture as easily as tissue-specific stem cells do. Human iPS cells offer advantages over human ES cells in that they do not begin with the destruction of a human embryo, which is a controversial process; iPS cells can also be made routinely from a patient's own cells, which should provide a better representation of a disease, especially if it is genetically based, and should not lead to any immune system response if those cells are used in the patient from which they were derived. However, iPS cells have only been in existence for a few years, and researchers are still assessing how well and stably they differentiate.

Turning back differentiation

Most of the excitement in stem cell research today centres on making iPS cells and taking their measure. Until a few years ago, most researchers believed that an individual cell and its progeny could only move from a primitive state to a more differentiated one. Going in the opposite direction was considered about as possible as travelling backward through time, and the skepticism became more entrenched when some studies indicating plasticity of adult-derived stem cells could not be readily reproduced.

By the early 2000s researchers had identified dozens of genes that are active in ES cells, and Yamanaka had an improbable idea: perhaps forcing differentiated cells to activate some combination of these genes would render the cells embryonic. The cloning of Dolly the sheep and other mammals had hinted that adult genomes could be reset by unknown components in oocytes13. From a practical perspective, such a project seemed daunting. There were many combinations of genes to test, and the most likely outcome was the completion of batteries of experiments without any interesting results. This turned out to be far from the truth. Yamanaka and his colleagues came up with a four-gene recipe for reprogramming differentiated mouse cells — and thus the first iPS cells were generated. Yamanaka, along with other researchers operating independently, worked to improve the process for making iPS cells and extended it to human cells.

Because of this, researchers have radically revised their ideas about the stability of the differentiated state. Indeed, the right combination of genes seems capable not only of making cells revert to a pluripotent state, but also, in some cases, of taking a cell directly from one specialized state to another14,15,16.

Improving the iPS cell

The first iPS cells were produced by inserting the 'pluripotency genes' (Oct4, Klf4, Sox2 and c-Myc) as part of a viral vector that integrated into the cells' genome. Amazingly, the inserted genes in the iPS cells shut themselves off as the reprogramming process was completed, and the cells reactivated their own pluripotency genes to maintain the pluripotent state. Also, the exact sites of insertion for the transgenes did not seem to matter. These observations hinted that reprogramming might be possible without needing to insert additional genes. What if it would be possible to just reactivate cells' own dormant pluripotency genes?

Having to use transgenes to obtain pluripotency is undesirable if the cells are destined for therapy, because of the potential risks of the process. How safe will the reprogrammed cells be? What if one of the inserted genes has knocked out an important tumour suppressor gene? The use of viral vectors means the resulting iPS cells, however well they perform, would be unlikely to be passed for medical use. Thus, 2008 and 2009 saw a race for new techniques to sidestep these problems. Viral vectors that did not integrate into the cells' genome (and whose DNA was rapidly degraded), as well as nonviral plasmid vectors, were shown to work, as were vectors that could insert themselves into the genome and then snip themselves out again when reprogramming was complete17,18,19,20. Other techniques supplemented genes with small molecules, whittling the genes required from four to three, then two21, and, at least in neural stem cells, to one22. In the spring of 2009, researchers finally showed that you could reprogram cells using small molecules and engineered proteins — without adding any introduced genetic material at all — albeit with very low success rates23,24.

The most rigorous assessments of iPS cells has been conducted on those generated through genetic engineering, which is the best characterized and most tractable technique. In July 2009, a new publication from Yamanaka's group suggested that iPS cells may need to be assessed particularly carefully25. Mouse iPS cell lines derived from some tissues, such as tail tips and liver, seem quite resistant to differentiation, growing vigorously into masses of undifferentiated cells even after being twice subjected to differentiation regimes. The implications of those findings are still unclear, however, because the cell lines in question differ in how the transgenes have integrated and so vary in other ways than just the tissue type from which they were derived. Another study of live-cell imaging in reprogramming human iPS cells found that some markers used to assess pluripotency can be misleading26.

What next?

The next few years will no doubt bring a torrent of publications highlighting the practical applications of iPS cells, and there are plenty of questions still to answer. Do different reprogramming techniques affect the resulting cells' differentiation potential and potential therapeutic usefulness? How does the originating tissue influence the cells' differentiation and pluripotency potential? How can iPS cells be efficiently converted to useful cell types such as neurons or pancreatic islets, either for personalized drug screening on cells from patients or for understanding basic processes in disease development? And, most importantly, what is the most expedient and accurate way to answer these questions?

Efficient and standardized ways of assessing the quality and potential of any given iPS cell line will be crucial for workers to be able to compare results from different labs. Now researchers can only assess how well human iPS cells have been reprogrammed by injecting them into an immunocompromised mouse and seeing how well they form a tumour known as a teratoma — a mass of various differentiated cell types. Such assays are time-consuming and can be hard to interpret. Expression profiles of genes and proteins, if they could be trusted, would be much more convenient. So far, however, a distinct signature of gene or protein expression that could be used to detect pluripotency has eluded researchers.

Cells differentiated from iPS cells have provided some therapeutic benefit in mouse models of sickle-cell anaemia, Parkinson's disease and cardiovascular disease27,28,29. Even so, there are big worries that transplanting cells derived from pluripotent stem cells (either ES or iPS cells) into patients would be premature. In a laboratory dish, pluripotent stem cells promise to become any sort of cell a researcher could want. But when transplanted into the body, such cells could grow unpredictably: clinicians worry most about tumour formation, although one can imagine other dangers, such as tissues growing in the wrong place — chunks of bone appearing in heart tissue, for example. Researchers are in the process of developing techniques to assess what risks cells derived from pluripotent cells pose for transplantation therapy, but such therapies are likely to be many years in the future.

A model of disease

Much earlier on the scene will be iPS cells used as models of disease. Researchers are frequently stymied because they simply cannot get enough material, such as tissues characteristic of a particular disease, to study. Animal models of disease are expensive and time-consuming to produce and use, and animals are capable of suffering and do not always reflect the human condition. But, more often than not, these models are scientists' only option.

One hope is that iPS cells will provide abundant sources of human tissue, allowing researchers to conduct more frequent, more relevant and less expensive studies of human disease. Many academic institutions have established facilities to reprogram cells from patients with particular chronic diseases. Cells taken from patients with Parkinson's disease19, Lou Gehrig's disease30, diabetes31 and familial dysautonomia32 have been reprogrammed to form iPS cells and then coaxed to form the types of cells affected by the disease. This could provide an unprecedented opportunity both to use human cells in culture to study disease progression and to use these cultures to screen new drug candidates. Such efforts are more than academic — both biotechnology start-ups and large pharmaceutical companies are going down this route.

But even to use iPS cells routinely for drug screening, researchers must solve a host of tedious problems. Arguably the simplest will be to establish stable supplies of pluripotent cell lines and make sure they perform consistently from batch to batch without accumulating problematic mutations or epigenetic modifications. More daunting problems occur when researchers cause iPS cells to become other cell types, which is a more complex process than keeping them undifferentiated in culture. So researchers need reliable means to differentiate sufficient numbers of cells into the desired type, to sort the desired cell types from the unwanted ones and to do so using techniques that are neither too time-consuming nor expensive. And the cells that are finally obtained must not only make markers characteristic of the desired cell type, but also safely and reliably mimic the cells' behaviour.

Cells used as models of disease for drug screening must be predictive of the drug's behaviour in the body. Cells that will be used to treat disease must meet different and more rigorous criteria: first, they must be able to arrive at and stay in their desired location; second, they must be able to persist and possibly proliferate within the patient, perhaps for a patient's entire life; third, they must evade the patient's immune system; and fourth, they must be able to function inside the patient, even if that means integrating into complex, damaged tissue like the substantia nigra in patients with Parkinson's disease or the heart muscle in those with cardiovascular disease. Most important, they must not pull any dangerous surprises: unexpected cells in the right place, expected cells in the wrong place or masses of cells growing uncontrollably anywhere in the body.

The natural version

Although new technologies for working with stem cells in the laboratory consistently generate headlines, the less publicized efforts to understand how stem cells work in the body are equally important. Stem cells persist in a variety of adult organs, and researchers are now uncovering some key signaling pathways governing their behaviour. Some of these pathways vary in different tissue types, but the themes are often similar. Epigenetic tags on DNA and chromatin proteins that direct cells to a particular fate have been uncovered in both neural and skin stem cells this year33,34,35,36,37. The way the strings of DNA in the chromosomes are coiled up and packaged seems particularly important for maintaining ES cells. Stem cells in muscle, brain and blood all change their regulatory pathways as organisms age: early on they are set to proliferate38,39, whereas later in life they tend to resist proliferation, perhaps to lower the risk of cancer, with aging as an inevitable side effect.

An understanding of what controls stem cells within the body will lead to ideas on how to manipulate them, and some experts believe drug regimens that affect a patient's own stem cells will be less risky and more powerful than delivering the cells themselves.

But such approaches could be a double-edged sword, affecting not just the target stem cells but others as well. Studies of cancer stem cells have shown that they rely on the same pathways as normal, healthy stem cells, suggesting that going after one could kill off the others40,41,42. On the other hand, some cancer stem cells seem to rely on embryonic or other pathways that are not used in healthy adult tissue43. Turning knowledge into medicine is never easy, but the pace of knowledge generation itself is fast and furious.

Related articles

Stem cells: Fast and furious

From skin cell to stem cell

Personalizing pluripotency

The Niche: Human reprogramming changes everything, and nothing

Race to mimic human embryonic stem cells

Cloning special: Dolly: a decade on

Smash the (cell) state

The Niche: Induced pluripotent stem cell reading list

Cell origin and variation in induced pluripotent stem cell lines

iPS cells used to probe rare degenerative disease

Cell therapy: Being patient

The architecture of pluripotency

Packaging DNA for pluripotency

Connecting Oct4 to chromatin

How tissue-specific stem cells keep their fate

Chromatin remodelling is essential for postnatal neurogenesis

Neural stem cells, young and old

A metasignalling network makes muscles age

How breast cancer resists treatment

Cancer stem cells resemble healthy stem cells, resist treatment

A screen for cancer killers

Heart, heal thyself

Mother of all heart cells

A sort of beta-cell magic: transdetermination seems easier than transdifferentiation

Diabetes stem-cell treatment looks to cell capsules

The cell that might save sight

Induced pluripotent stem cells, down to one factor